The skills shortage in high-performance computing (HPC) is an urgent issue that needs to be addressed.
The HPC market has huge potential not just for individual businesses that can use the technology to drive profitability and business advantage, but also for the broader economy where it could drive up GDP.
Without the right skill-sets, however, this potential is likely to remain unfulfilled. After all, there is no point buying the fastest car if you haven’t first learnt how to drive it.
“It’s not enough to keep building powerful supercomputers unless we have the brains,” said Stan Ahalt, a 20-year supercomputing veteran and director of a supercomputing centre with its headquarters at the University of North Carolina at Chapel Hill. “Think of a supercomputer as a very fast racing engine. We need more drivers to use those engines.”
>See also: Two thirds of high performance computing sites use big data – IDC
As a nation too, we need a body of application developers who understand multi-core architectures and how to scale applications. If we don’t develop these people, we are likely to fall behind our competitors.
Of course, there’s no one single answer here. More often than not, the skills to support HPC are so sought after that companies are relying on expensive consultancy to hand-hold them through the various implementation stages.
The industry needs to help plug this skills gap to capture the best knowledge sets and connect the users with consultants out in the field.
The government has already played its part – not just through its 2011 decision to pump £145 million into supercomputing in the UK, but also through its support for specific projects including funding centres for doctoral training in leading universities, accessing national e-infrastructure.
Technology can play a key role also in helping to plug the skills gap. To be successful in the future, HPC needs to move away from a focus on ‘closed-in’ proprietary architectures towards a more open standards-based approach, making it easier for less specialist staff to use systems and solutions.
Currently, HPC technology is still trying to catch up with the kind of standardisation seen elsewhere in the ICT space.
To accelerate adoption and help bridge the skills gap at the same time, vendors first need to make systems easy to understand and use. This will help drive their customers to adopt scalable and flexible architectures and make the technology more accessible to their user base.
Secondly, providers need to be prepared to work with their commercial customers to ensure that their software and applications are modified to be capable of leveraging the size and scale of computing systems that HPC typically has to support.
As a result, we are seeing growing momentum behind a new approach, HPC-on-demand (or HPC clouds) – allowing businesses to access HPC resources as and when required through an online portal.
By using this approach, companies no longer need to worry about the complexities of running their own environment, or the need for highly skilled specialised teams. Instead, they can tap into the available computing capability across the web as and when needed.
The emergence of HPC-on-demand is opening up the technology to a wider range of companies with basic computing resources and a lack of specialist IT skills.
Freed from the shackles of complex technology, such organisations now have freedom to innovate. With HPC-on-demand, they have no need for physical infrastructures and no more maintenance requirements, while still having access to high-performance tools for innovation.
>See also: UK nuclear-warhead maker deploys three distributed-memory supercomputers
The government’s willingness to provide financial support for the development of HPC in the UK is to be applauded. But money will not be sufficient in itself to address the skills shortages and drive the growth of the sector.
Overcoming the serious skills shortage the country is currently facing requires a potent mix of talent and technology. If UK PLC is to benefit fully from the potential of this exciting new market, both of these key elements need to be properly harnessed, nurtured and developed.
Sourced from Andy Grant, director, HPC and big data practice, Bull UK & Ireland
